From cf4dc28c5fa762c93dcc2fd51d42a7bb684f7ecb Mon Sep 17 00:00:00 2001
From: E Rasche <hxr@hx42.org>
Date: Tue, 1 Aug 2017 15:15:17 +0200
Subject: [PATCH] Disable sanitization by default

---
 scripts/grt.py         | 10 +++++++---
 scripts/grt.yml.sample |  7 ++++++-
 2 files changed, 13 insertions(+), 4 deletions(-)

diff --git a/scripts/grt.py b/scripts/grt.py
index af2e5a27ed8..1be4f080c95 100644
--- a/scripts/grt.py
+++ b/scripts/grt.py
@@ -198,6 +198,7 @@ def main(argv):
     CHECK_POINT_FILE = os.path.join(REPORT_DIR, '.checkpoint')
     REPORT_IDENTIFIER = str(time.time())
     REPORT_BASE = os.path.join(REPORT_DIR, REPORT_IDENTIFIER)
+    SANITIZATION_ENABLED = config['sanitization']['enabled']
 
     if os.path.exists(CHECK_POINT_FILE):
         with open(CHECK_POINT_FILE, 'r') as handle:
@@ -219,7 +220,7 @@ def main(argv):
     job_state_data = defaultdict(int)
 
     annotate('san_init', 'Building Sanitizer')
-    san = Sanitization(config['blacklist'], model, sa_session)
+    san = Sanitization(config['sanitization'], model, sa_session)
     annotate('san_end')
 
     if not os.path.exists(REPORT_DIR):
@@ -313,8 +314,11 @@ def main(argv):
                 .filter(model.JobParameter.job_id <= min(end_job_id, offset_start + args.batch_size)) \
                 .all():
 
-            unsanitized = {param[1]: json.loads(param[2])}
-            sanitized = san.sanitize_data(job_tool_map[param[0]], unsanitized)
+            if SANITIZATION_ENABLED:
+                unsanitized = {param[1]: json.loads(param[2])}
+                sanitized = san.sanitize_data(job_tool_map[param[0]], unsanitized)
+            else:
+                sanitized = param[2]
 
             handle_params.write(str(param[0]))
             handle_params.write('\t')
diff --git a/scripts/grt.yml.sample b/scripts/grt.yml.sample
index 7ed7ca9e8db..622fb155b46 100644
--- a/scripts/grt.yml.sample
+++ b/scripts/grt.yml.sample
@@ -16,7 +16,12 @@ grt:
     share_toolbox: True
 
 
-blacklist:
+sanitization:
+    # This defaults to disabled as it has a serious performance impact and may
+    # not be necessary for your instance. Without sanitization we see
+    # performance on the order of 4k parameters parsed per second. With
+    # sanitizatoin on, it averages to 100 parameters per second.
+    enabled: False
     # Blacklist the entire tool from appearing
     tools:
         - __SET_METADATA__